Spec: Add pagination pattern documentation

[a6b8]

Problem

Many APIs use pagination (offset/limit, cursor-based, page-based) but FlowMCP v2.0.0 has no documented best practice for how schemas should handle paginated endpoints. Schema authors currently reinvent this for every schema.

Examples from Schema Research

After reviewing 300+ APIs, pagination patterns appear in roughly 60% of data-heavy APIs:

Pattern	APIs Using It	Example
Offset/Limit	GBIF, OpenFDA, FRED, World Bank	offset=0&limit=100
Cursor-based	GitHub, Discogs, iNaturalist	cursor=abc123
Page/PerPage	Europeana, MusicBrainz, OMDb	page=1&per_page=25
Token-based	NASA TechPort, Google APIs	pageToken=xyz
Start/Count	CoinGecko, DeFi Llama	start=0&count=50

Current State

Schema authors expose pagination parameters as regular position parameters (query params), which works technically but:

• No consistent naming convention across schemas
• No guidance on default values
• No documentation on how agents should handle multi-page results

Proposed Addition

Add a cookbook/best-practice section to the spec covering:

1. Recommended parameter names for each pagination pattern
2. Default values — what defaults make sense for AI agent consumption
3. Schema-level prompt template for paginated endpoints (once prompts are available)
4. Examples for offset/limit, cursor, and page-based patterns

Priority

P3 — Quality-of-life improvement for schema authors and consistency across the ecosystem.